Mixed Mode Device-to-Device Communication Scheme for ...
Transcript of Mixed Mode Device-to-Device Communication Scheme for ...
Tanzania Journal of Science 45(3): 405-416, 2019 ISSN 0856-1761, e-ISSN 2507-7961
Β© College of Natural and Applied Sciences, University of Dar es Salaam, 2019
405 http://journals.udsm.ac.tz/index.php/tjs www.ajol.info/index.php/tjs/
Mixed Mode Device-to-Device Communication Scheme for Congestion
Reduction and Channel Usage Optimization in 5G Cellular Networks
Chiza M Christophe*, Omar F Hamad, Libe V Massawe, Abdi T Abdalla
Department of Electronics and Telecommunications Engineering, College of Information and
Communication Technologies, University of Dar es Salaam, Tanzania. *Corresponding author e-mail: [email protected];
Co-authors e-mail addresses: [email protected], [email protected], [email protected]
Abstract Device-to-Device (D2D) communication schemes have gained more attention in cellular
networks particularly in normalization process of the upcoming 5G networks. They have been
investigated in core network offloading, congestion reduction and channel usage optimization.
The two last cases are among the major constraints in current cellular networks and are the main
concerns of this paper. The paper presents a mixed mode D2D communication scheme to
decentralize data collection between devices and the base station in order to reduce the number
of direct connections at the base station of ultra-dense cells characterized by different levels of
channel utilizations or target data rates, as expected for 5G networks. The attachment utility is
derived as the overall gain of a device for a target data rate and is used as a metric for D2D
associationβs decision. Results show that the attachment utility and D2D pairs increase by either
increasing the D2D communication range or decreasing devicesβ target data rates. A further
important consideration is that the proposed mixed mode D2D communication scheme improves
the throughput expectation in the cell by 14.2% compared to the regular cellular communication.
Keywords: 5G Networks, Channel Usage Optimisation, Congestion Reduction, D2D
Communication Scheme, Target Data Rate.
Introduction
Current mobile communication
infrastructures are expected to experience
overloading due to overgrowing data traffic
which demands extremely high data rates (Hu
et al. 2019). Therefore, mobile communication
pioneers have been subjected to an urgent need
of developing enhanced mechanisms and
features to cope with the current and the
expected future mobile traffic overloading. The
upcoming 5G networks will provide
meaningful advantages to mobile network
operators (MNOs). The 5G networks are
expected not only to support the very high data
transmission demand wanted by current users,
but also bring an assurance in investigating and
creating additional services and applications
(Singh and Chawla 2017). Compared to classic
infrastructure based cellular communications,
5G networks will carry different modes of
communications in a single cell and enhance
network slicing allowing on demand and
flexible radio resource allocation (Lee et al.
2019, Sattar and Matrawy 2019). Multiple
communication modes lead to a high network
throughput and overhead reduction at the
access point or base station (Sheybani et al.
2018, Lee and Lee 2019). With multiple modes
of communications, devices operate either in
regular, D2D or both modes based on their
locations and required quality of service (Lee
and Lee 2019).
The Device-to-Device (D2D)
communication is viewed as a potential
component of the upcoming 5G network
infrastructures because it enables devices
which are closer to initiate direct
communication rather than interacting via the
Christophe et al. - Mixed mode device-to-device communication scheme β¦
406
base station or remote radio head (Kar and
Sanyal 2017, Li et al. 2019). It was introduced
by the 3GPP to support proximity services
(ProSe) in enhancing network performance or
enabling communications in circumstances
where network infrastructures may not be
operational to support information and alerts
exchange; circumstances such as terror
situations, tsunami and earthquakes (Kunz et
al. 2013, Jung and Kim 2016). In addition, the
emergence of multimedia services has
triggered the integration of D2D
communication in LTE Release 12 networks to
enable content sharing when subscribers or
devices that are closer to each other request for
the same content (Liu et al. 2012, Gupta et al.
2018, Seth and Sharma 2018).
With the promising consideration of D2D
communication, cooperative D2D
communication and relay assisted D2D
communication have indeed gained more
attention (Lee and Lee 2019). These two
concepts are used to extend the cell coverage,
to improve the transmission reliability, to
reduce congestions, to reduce the power
consumption, and to upgrade either the
throughput or load balancing in the core
network (Jung and Kim 2016, Li and Cai 2018,
Gao et al. 2019). In cooperative D2D
communication, devices interactively collect
their contents at one point and use a single link
to reach the base in order to reduce congestions
and the feedback load at the base station (Gui
and Deng 2018, Li and Cai 2018, Lee and Lee
2019). In relay assisted D2D communication, a
user equipment (UE) at the edge of a cell
attaches at its serving base station by relaying
its content through a UE to UE communication.
This operation reduces the power consumption
while improving the transmission efficiency
(Qiao et al. 2010, Jung and Kim 2016, Lee and
Lee 2019). Due to the challenge of the first
nearest relay unavailability, the research done
by Rajabi and Ghorashi (2018) has studied the
impact of connecting to the nth
nearest device
rather than connecting to the first nearest
device. Indeed, D2D load balancing, resource
allocation and routing algorithms have been
provided by Zhang et al. (2018) such that a
device at the edge of a congested small cell
attaches to the closest uncongested small cell
by using devices in the same path as relays.
Studies on relay assisted D2D
communication have, however, investigated
ideal environment devices which fully utilize
the allocated channels (Qiao et al. 2010, Jung
and Kim 2016, Rajabi and Ghorashi 2018).
Therefore, with the ultrahigh 5G channel
capacity that can accurately support multiple
devices, a regular device can be allocated a
full channel although its target data rate
requires the portion of the channel capacity
(Jian et al. 2015). Consequently, a device
serving as a relay can attach data when the
served device does not fully utilize the
allocated channel, but the investigated relay
assistance scenarios do not enable a relay to
attach its content.
The main contribution of this paper is
providing analysis of mixed mode D2D
communication taking into account the target
data rate in congestion reduction and channel
usage optimisation. Different from previous
works which considered that devices fully
utilize the allocated channels and focused on
the relay assistance aspect of devices, this
paper presents the feasibility of data
aggregation in ultra-dense cells characterized
by different levels of channels utilization or
target data rates. Both the D2D transmission
and the regular transmission are considered,
and generate a mixed mode D2D transmission.
With this generated transmission mode,
analysis is done such that two devices
aggregate their data by using D2D mode and
use a common regular cellular link to transmit
the collected data to the base station based on
their target data rates. This procedure is
envisioned because a real environment device
can partly utilize the allocated channel while its
neighbours have missed channels.
Materials and Methods
Overview of the analysis
This paper presents the obtained
simulation results in form of plots, and
Tanz. J. Sci. Vol. 45(3), 2019
407
discussions in form of insights. The software
material used for numerical implementation
of equations is MATLAB simulator.
Numerical simulations were run by varying
various parameters, mainly the number of
devices in the cell coverage area, the D2D
communication range and the target data
rates. Statistical probabilistic analysis is used
as the method for modelling the randomized
distribution and location of devices in the cell
coverage area, as in Rajabi and Ghorashi
(2018), Sheybani et al. (2018), Lee and Lee
(2019). The channel utilization is done based
on the target data rate and the rest of the
channel capacity should be available for other
devices. Hence, accessing the reserved
channel is opportunistic and has to be
modelled probabilistically.
The paper presents opportunistic
attachment when devices in the same cell
coverage area do not fully utilize the allocated
channels. The scenario is mathematically
modelled such that devices follow the Poisson
Point Process (PPP) with different
distribution densities, and independently
utilize the allocated channels. The probability
at which devices are located in each other
D2D communication range is related to the
probability at which Poisson points fall in a
given area. This probability is called falling
probability throughout this paper. Indeed, the
probability at which two devices can share the
same channel without exceeding the
maximum channel capacity defines matching
probability. These probabilities are combined
to give the attachment utility that represents
the overall gain of a device for a target data
rate, and characterizes the chance at which the
device can aggregate its content with a
neighbour that does not fully utilize the
allocated channel. In some cases, multiple
devices are assumed nearby the same
neighbour which does not fully utilize the
allocated channel, these devices compete by
tuning their target data rates in order to
improve their attachment utilities. A high
attachment utility means that a device is
likely to get attached and trigger the data
aggregation process. The target data rate
denotes the portion of the maximum channel
capacity which satisfies the deviceβs need.
In the analysis, the base station is located
at the centre of the cell and defines the cell
coverage area in which devices are randomly
distributed. The base station has less number
of channels or wireless resources than
required. Therefore, as illustrated in Figure 1,
some devices will be granted the available
communication channels (Type I devices) and
others will miss (Type II devices). The
variables π and π, respectively, represent
the number of Type I and Type II devices in
the cell coverage area. So, given a particular
device with D2D communication range π , the
D2D communication coverage is ππ 2. Assuming that a Type I device fully or
partially utilize the allocated channel
capacity, a Type II device can be attached to a
near Type I device by using the D2D mode as
long as the maximum channel capacity is not
exceeded. Based on the competition and
complexity aspects of this scenario, the
analysis is carried out in different
environments such as simple homogeneous,
simple heterogeneous, and complex
homogeneous.
Simple homogeneous environment Consider a simple homogeneous environment
(S.Ho.Env) to be a cell with π randomly
distributed Type I devices and one Type II or a
Particular device, as in Figure 1. Also a cell in
which each Type I device falling in the Type II
deviceβs coverage area supports the attachment.
The probability of having at least π£ points
(transmitters) in a given subset (π΄) of the cell
coverage area has been provided in Rajabi and
Ghorashi (2018). This probability is given by
ππ£ in equation 1.
ππ£ = π(π β₯ π£, π΄) = 1 β β(ππ΅Γπ΄)π
π!πβππ΅Γπ΄.π£β1
π = 0 (1)
Christophe et al. - Mixed mode device-to-device communication scheme β¦
408
The variables π and ππ΅ represent the number
of points in the subset π΄ and the distribution
density of points in the cell coverage area,
respectively. From equation 1, the probability
of having at least one Type I device in a D2D
communication area (π΄) under assumption
that the cell coverage area contains π Type I
devices, is expressed in equation 2.
π1 = π(π β₯ 1, π΄)
= 1 β(ππ Γ π΄)π
π!πβππΓπ΄β π = 0
= β(ππ Γ π΄)π
π!πβππΓπ΄.
π
π = 1
(2)
Figure 1: Distribution of devices in the cell coverage area.
In this case, ππ represents the distribution
density of Type I devices in the cell coverage
area. For simplicity, let us denote π1 in
equation 2 to yield equation 3;
π1 = β ππ[π|π΄],
π
π = 1
(3)
where ππ[π|π΄] represents the falling
probability or the probability at which π
Type I devices fall in a given coverage area π΄.
To fit with the Poisson Point Process (PPP)
used in Rajabi and Ghorashi (2018),
ππ [π| π΄] is constrained such that π β0,1,2, β¦ . π and
0 < β ππ[π|π΄]
π
π = 0
β€ 1.
(4)
With this constraint in equation 4,
1β ππ[0| π΄] holds the Choquet capacity of the
PPP in Jeulin (2014).
Considering that in a S.Ho.Env each Type
I device falling in π΄ supports the attachment
at 100%, the matching probability (πππ»π)
between a Type II device and a ππ‘β Type I
device falling in π΄ is equal to one. The sum of
individual matching probabilities is π for π
Type I devices in π΄. When a cell embeds π
Type I devices and a part of them randomly
falls in π΄, the matching and falling
probabilities are multiplied to obtain the
attachment utility π given by equation 5.
π = β (ππ[π| π΄] β πππ»π
π
π = 1
)
π
π = 1
= β π Γ ππ[π|π΄].
π
π = 1
(5)
Equation 5 considers that each Type I device
supports the attachment. However, in real
environment different devices utilize the
allocated channels independently. For
instance, a device which needs to fully utilize
the allocated channel cannot support the
attachment, unless they cooperate to adjust
their target data rates. The difference and
independence in channel utilization generate a
sort of heterogeneity. Hence, a simple
heterogeneous environment is studied as in
subsection below.
Tanz. J. Sci. Vol. 45(3), 2019
409
Simple heterogeneous environment
Consider a simple heterogeneous environment
(S.He.Env) as a cell that contains π Type I
devices and one Type II device. In this case,
each device is supposed to utilize the
allocated channel differently and
independently. Based on this assumption, the
attachment is conditioned by the probability
of having π Type I devices in π΄ and the
probability at which the sum of target data
rates do not exceed the maximum channel
capacity. To analyze the impact of these two
probabilities on the attachment utility, the
notion of target data rate or the percentage of
the maximum channel data rate which
satisfies a deviceβs needs is introduced. Thus,
let π» = {β1, β2, β3 ,β¦β¦,βπ } with π = 1, 2, β¦ π,
represent a set of linearly distributed data rate
percentages, and π = {π1, π2, π3 ,β¦β¦,ππ } be an
equivalent set of data rates. Negative and zero
values are not assumed as possible target data
rates. Therefore, the value of ππ is defined
in ] 0, πΆ ], a subset in the conventional set of
positive nonzero real numbers, where πΆ
represents the maximum channel capacity or
regular cellular link data rate.
For ππ πππ ππ β π, where ππ and ππ are
target data rates of the mth
Type I device and
the nth
Type II device, a Type I-Type II
association can be enabled for a cell channel
capacity πΆ if:
(ππ + ππ) Γ 1
1 β Β΅ β€ πΆ. (6)
The component Β΅ represents the data
collection time, and is constrained such that 0
β€ Β΅ < 1 second when the data rate is
expressed in π₯bit/sec. The variable π₯ is G
when the data rate is expressed in Gigabits
per second (Gbit/sec) or M for megabits per
second (Mbit/sec). Equation 6 is used when
the collection time is included in the
transmission time frame. When devices
collect and format data before the
transmission time occurs, the collection time
is excluded in the transmission time and then
equation 6 reduces to ππ + ππ β€ πΆ. When
the Type II device targets a data rate ππ at ith
position in the set Q, and Type I devices
target their data rates in the same set, the total
number of combinations for ππ with each
element of the set Q such that ππ + ππ β€ πΆ
is held, is equals to πΏ = π β π. Therefore, the
matching probability in S.He.Env is given by
πππ»π in equation 7.
πππ»π =
πΏ
π =
π β π
π. (7)
Combining equations 4 and 7, the attachment
utility (ππ·) in S.He.Env is expressed as a sum
of individual falling probabilities and
matching probabilities products. This is given
by equation 8;
ππ· = β (ππ[π| π΄] β πππ»π
π
π = 1
) ,
π
π = 1
(8)
where ππ· represents the attachment utility for
a given Type II device π· with a D2D
coverage area π΄ and a target data rate at the ith
position in the set Q. πππ»π represents the
probability at which π· can match with the ππ‘β
Type I device falling into its coverage area π΄.
The variable π β {1, 2, β¦ , π} when π Type I
devices fall in π΄. As all the target data rates
are independently chosen in the same set π, the same matching probability is applied such
that:
π1π»π = π2
π»π = π3π»π = β¦ = ππ
π»π
= π β π
π.
Then, equation 8 becomes
ππ· = π β π
πβ π Γ ππ[π| π΄]
π
π = 1
.
(9)
The attachment utility ππ· in equation 9 is
applied when a cell undelays π Type I
devices and a single Type II device. Also, the
equation is applicable when there are π Type
II devices in the cell coverage area and
individual Type II devicesβ coverage areas are
disjoint or not overlapping each other. The
two scenarios above are ideal because they
occur rarely in real environment.
Furthermore, when different Type II devices
overlap each otherβs coverage areas, they
conflict, the system becomes complex and
Christophe et al. - Mixed mode device-to-device communication scheme β¦
410
analysis imposes additional assumptions.
Therefore, the attachment utility in a complex
homogeneous environment (C.Ho.Env) is
derived as in the following subsection.
Complex homogeneous environment
This scenario characterizes a cell with π
Type I and π Type II devices, the devicesβ
coverage areas may overlap each other, and
each Type II device can attach to any nearest
Type I device. Any Type II device influences
the attachment utility of a particular device π·
if it is either in the coverage area π΄ or it is in
a coverage area of radius 2π taking the
position of π· as the centre or reference
position. Therefore, all devices in the
coverage area equivalent to π = π(2π )2
impact the attachment utility of π·. The
probability π1π at which at least one Type II
device is located in π is expressed in equation
10 by recalling π1 from equation 3:
π1π = β ππ[π| π]
π
π = 1
. (10)
Up to this point, all the derived attachment
utility equations consider that π· is alone in
the cell. Therefore, its attachment utility is
maximum, but when different Type II devices
conflict with π·, the utility is eventually
shared. For the sake of compatibility with the
homogeneity concept, the attachment utility is
found such that equation 5 is shared equally.
Thus, if π β 1 Type II devices conflict
with π·, the attachment utility of π· is given
by ππ·π. ππ·π = 1
ππ , with π = 1, 2, β― , π.
Equivalently, ππ·π can be rewritten as in
equation 11.
ππ·π = [1 β (π β 1
π)] β π
π
π = 1
Γ ππ[π| π΄].
(11)
The component (πβ1
π) represents the impact
of conflicting Type II devices on the
attachment utility, and it is zero when π· is
alone (π = 1) in π. Besides, the impact of
(πβ1
π) is influenced by the probability of
having π devices in π. Therefore, the utility
ππ·π can be rewritten as an optimized utility
ππ·π,π by introducing a term βπ , dependent of
π and the falling probability of π, such as in
equation 12.
ππ·π,π = [1 ββπ] β π Γ ππ[π| π΄],
π
π = 1
(12)
where βπ = (πβ1
π) ππ[π| π].
As assumed, the base station coverage area
contains π Type II devices; they can partially
or totally fall into π. Thus, considering the
variability of π, the average attachment utility,
ππ·π,πππ£ of a Type II device is expressed in
equation 13.
ππ·π,πππ£ = ( β π Γ ππ[π| π΄]
π
π = 1
)
Γ1
πβ [1 ββπ]
π
π = 1
.
(13)
Suppose devices are randomly distributed in
the cell coverage area by following a PPP, with
two different distribution densities. With this
assumption, ππ[π| π΄] and ππ[π| π] are replaced
by their probability density functions, and
equations 5, 9 and 13 yield to equations 14, 15
and 16, respectively.
π = β(ππ Γ π΄)π
(π β 1)!πβππΓπ΄,
π
π = 1
(14)
ππ· = π β π
πβ
(ππ Γ π΄)π
(π β 1)!πβππΓπ΄,
π
π = 1
(15)
ππ·π,πππ£ = ( β
(ππ Γ π΄)π
(π β 1)!πβππΓπ΄
π
π = 1
) Γ1
πβ [1 β (
π β 1
π)
(ππ Γ π)π
π!πβππΓπ]
π
π = 1
, (16)
Tanz. J. Sci. Vol. 45(3), 2019
411
where ππ = π
πΓ(π π΅)2 and ππ =
π
πΓ(π π΅)2 are the distribution densities of Type I and II devices,
respectively.
Results and Discussions
Numerical simulations were run in
MATLAB by varying the number of Type I
and II devices in the cell coverage area, the
D2D communication range π and devicesβ
target data rates. Equations 14, 15 and 16
were implemented in MATLAB by
considering a dense 5G micro cell with a
radius π π΅ = 1000 π and a matrix π― (1 Γ10) was constructed to get a linearly
distributed set of data rate percentages and a
1Gbps maximum channel capacity for simple
heterogeneous environment. These three
equations represent the attachment utility in
simple homogeneous, simple heterogeneous
and complex homogeneous environments
under Poisson Point Process, respectively.
Results in Figure 2 represent the overall
impact of the number of Type I devices on the
attachment utility under different
environments. It is observed that the
attachment utility increases with the number
of Type I devices. This means that, the larger
the π, the more a device is likely to find a
Type I as its neighbour. The attachment
utility in simple homogeneous environment
outperforms utilities in other environments,
this is because it constitutes the upper bound
attachment utility where each Type I device
has to support the attachment, and Type II
devicesβ coverage areas are not supposed to
overlap or conflict each other.
Figure 2: Impact of the number of Type I devices on the attachment utility.
When devices are targeting small
percentages of the maximum channel
capacity, the attachment utility in simple
heterogeneous environment is getting closer
to the simple homogeneous environment, and
this is zero when the entire channel capacity
Christophe et al. - Mixed mode device-to-device communication scheme β¦
412
is assumed. A total of 1000 trials Monte Carlo
simulations of the simple heterogeneous
environment (M.C.S of S.He.Env) have been
performed for fair comparison between the
simple homogeneous and simple
heterogeneous environments. Results from
M.C.S of S.He.Env show that the generalized
attachment utility in a simple heterogeneous
environment is less than the attachment utility
of a device which targets 40% of the
maximum channel capacity and it is around
45% of the upper bound attachment utility.
This aspect elucidates the constraint of the
matching probability on the attachment
utility.
Figure 3 shows the variation of the
attachment utility with the D2D
communication range. It is observed that the
increase in D2D communication range
improves the attachment utility. The
improvement is based on the aspect that
devices are randomly distributed in the cell
coverage area and they can be located at any
position in the cell. Therefore, a large D2D
coverage area gives the potentiality of
embedding a large number of Type I devices,
this in return increases the attachment utility.
Furthermore, the attachment utility in a
complex homogeneous environment is quasi
constant with respect to the number of
conflicting Type II devices. With the
attachment utility expressed in terms of
average as in equation 13, the impact of a
large number of conflicting Type II devices
on the averaged attachment utility is
minimized by a small falling probability.
Therefore, graphs in Figure 3 appear straight
combined for different values of conflicting
Type II devices.
Figure 3: Impact of the D2D communication range on the attachment utility.
Tanz. J. Sci. Vol. 45(3), 2019
413
Figure 4 illustrates the impact of the
number of Type II devices on the attachment
utility for different D2D communication
ranges. As the D2D communication range
increases, the attachment utility also
increases. The attachment utility decreases at
a small scale with respect to the number of
conflicting Type II devices. However, above a
certain value of π, the attachment utility
increases as in Figure 4 (b), (c) and (d). This
is due to the fact that the impact of conflicting
Type II devices and the likelihood at which
they fall into π are varying at different scales.
In other words, for a certain value of π,
the influence starts decreasing and thus, the
attachment utility starts increasing. The
impact of conflicting devices increases
linearly with π while the falling likelihood
decreases exponentially with π, this refers to
the component βπ in equation 12. For small
values of π, the impact of conflicting devices
and their falling probability as given by βπ
are closer to each other, and hence the
increase of π results into the decrease of the
attachment utility, for example, when the
number Type II devices is less than or equal
to 50 in Figure 4 (b) and less than or equal to
20 in Figure 4 (c). However for large values
of π or when the number Type II devices is
greater than 50 in Figure 4 (b) and greater
than 20 in Figure 4 (c), the impact of
conflicting devices and their falling
probability mismatch, the falling probability
significantly decreases and hence the
attachment utility starts increasing.
Figure 4: Impact of N Type II devices on the attachment utility in complex homogeneous
environment.
The sum throughput represents the
aggregate throughput of all attached devices
in the cell (Pradhan et al. 2018). In this
paper, the sum throughput is estimated by
the sum target data rate of devices which are
supported by the base station. In regular
cellular communication the sum throughput
is therefore the sum of regular devicesβ
target data rates. In the proposed mixed
mode D2D communication the sum
Christophe et al. - Mixed mode device-to-device communication scheme β¦
414
throughput sums both regular devicesβ
target data rates and the D2D
communication devicesβ target data rates.
The D2D association rate in Figure 5
illustrates the extent to which D2D
associations are performed in the
investigated mixed D2D communication
whereas a single cellular link is assumed for
a pair of associated devices instead of a
specific cellular link for each device in order
to reduce the number of direct connection at
the base station. This figure shows the
number of created D2D pairs based on
complex homogeneous environment
scenario. It is observed that the rate of
association increases on average by 15.9%
when the D2D communication range is
doubled.
Figure 5: D2D association rate in the proposed mixed mode D2D communication.
Also the throughput expectation increases in
the cell with mixed mode D2D
communication compared to the regular
cellular communication scenario. During
simulations, the sum throughput was
randomly varying in single simulation trial.
This led to the use of Monte Carlo
simulations that estimate a random varying
entry through repeated experiments. With
1000 Monte Carlo simulations trials, it is
observed in Figure 6 that the sum
throughput of proposed mixed mode D2D
communication outperforms the regular
cellular communication on average by
14.2% when the same number of
communication channels is assumed for the
two communication modes. The
outperformance is due to the channel
sharing consideration that optimizes the
channels usage probability and hence
increases the sum throughput in mixed
mode D2D communication. In addition, the
throughput expectation improves in the cell
by 4.9% when the D2D communication
range is doubled.
Figure 6: Sum throughput estimation in the micro cell.
Tanz. J. Sci. Vol. 45(3), 2019
415
Conclusions
This paper proposes a statistical probabilistic
analysis of opportunistic attachment in mixed
mode D2D communication based on 5G
network expectations. Results provide clear
evidence that devices which missed the
attachment can participate in mixed mode
transmission by adjusting their utilities
through target data rate tuning. In other
words, a device which missed the channel is
likely to aggregate its content with a near
device which does not fully utilize the
allocated channel when it reduces its target
data rate. Also, its chances increase when the
D2D communication range is increased in the
cell. It is observed that the D2D
communication range has a considerable
impact on the attachment utility and the
number of D2D pairs. Therefore, its
normalization is of great interest to permit the
adjustment of the attachment utility in order
to attach a balanced number of devices. The
D2D communication range normalization,
however, remains an open research problem.
Simulation results show that the proposed
mixed mode D2D communication improves
the throughput expectation in the cell by
14.2% compared to the conventional regular
communication. Investigations for the utility
in complex heterogeneous environment and
channel state information are not presented.
These additional gaps remain open problems
and will inspire our future investigations.
References
Gao Y, Xiao Y, Wu M, Xiao M and Shao J
2019 Dynamic social-aware peer selection
for cooperative relay management with
D2D communications. IEEE Trans.
Comm. 67: 3124-3139.
Gui J and Deng J 2018 Multi-hop relay-aided
underlay D2D communications for
improving cellular coverage quality.
IEEEAccess 6: 14318-14338.
Gupta S, Trivedi A and Pawar P 2018 An
approach for improving performance of
underlay D2D communication. IEEE
Conference on Information and
Communication Technology, Jabalpur I: 1-
5.
Hu Y, Li X, Chen Y and Zeng B 2019 Multi-
principal multi-agent contract designβ―:
Theory and application to device-to-device
communications. IEEE Trans. Vehic.
Technol. 68: 2535-2544.
Jeulin D 2014 Introduction to some basic
random morphological models. Stochastic
Geometry Spatial Statistics and Random
Fields, hal-00770234: 978-3-319-10063-0.
Jian Q , Shen X, Mark W, Shen Q, He Y, and
Lei L 2015 Enabling device-to-device
communications in millimeter-wave 5G
cellular networks. IEEE Comm. Mag. 53:
209-215.
Jung S and Kim J 2016 A new way of
extending network coverageβ―: Relay-
assisted D2D communications in 3GPP.
ICT Express 2: 117-121.
Kar UN and Sanyal DK 2017 An overview of
device-to-device communication in
cellular networks. ICT Express 4: 203-208.
Kunz A, Tanaka I and Husain S 2013 Disaster
response in 3GPP mobile networks. IEEE
International Conference on
Communications Workshops Budapest:
1226-1231.
Lee J and Lee JH 2019 Performance analysis
and resource allocation for cooperative
D2D communication in cellular networks
with multiple D2D pairs. IEEE Comm.
Lett. 23: 909-912.
Lee W, Na T and Kim J 2019 How to create a
network slice? - A 5G core network
perspective. 2019 21st International
Conference on Advanced Communication
Technology PyeongChan: 616-619.
Li N, Lei G, Manogaran G and
Mavromoustakis X 2019 D2D
communication mode selection and
resource optimization algorithm with
optimal throughput in 5G network. IEEE
Access 7: 25263-25273.
Li Y and Cai L 2018 Cooperative device-to-
device communication for uplink
transmission in cellular system. IEEE
Transactions on Wireless
Christophe et al. - Mixed mode device-to-device communication scheme β¦
416
Communications 17(6): 3903-3917.
Liu Z, Peng T, Xiang S, Wang W 2012 Mode
selection for Device-to-Device (D2D)
communication under LTE-Advanced
networks. IEEE International Conference
on Communications Ottawa: 5563-5567.
Pradhan A, Basu S, Sarkar S, Mitra S, Roy SD
2018 Implementation of relay hopper
model for reliable communication of IoT
devices in LTE environment through D2D
link. 10th
International Conference on
Communication Systems and Networks
Bengaluru: 569-572.
Qiao J, Cai X and Shen S 2010 Multi-hop
concurrent transmission in millimeter
Wave WPANs with directional antenna.
IEEE ICC. 2010: 1-5.
Rajabi S and Ghorashi A 2018 Impact of
connecting to the nth
nearest node in
dedicated device-to-device
communications. Electron. Lett. 54: 535-
537.
Sattar D and Matrawy A 2019 Optimal slice
allocation in 5G core networks. IEEE
Networking Letters 1(2): 48-51.
Seth A and Sharma A 2018 Evolved
multimedia broadcasting and multicasting
services in LTE-A using Device to Device
communication. IEEE International
Conference on Inventive Research in
Computing Applications Coimbatore:
1250-1255.
Sheybani M, Mehrjoo M and Kazeminia M
2018 A proximity mode selection method
in Device to Device communications.
IEEE International Conference on
Computer and Knowledge Engineering 8
(Mashhad): 69-72.
Singh S and Chawla M 2017 A review on
millimeter wave communication and
effects on 5G systems. IARJSET: Int.
Adv. Res. J. Sci. Engin. Technol. 4: 28-
33.
Zhang H, Song L, Jun Y and Zhang A 2018
Load balancing for 5G ultra-dense
networks using device-to-device
communications. IEEE Tans. Wire.
Comm. 17 (6): 4039-4050.